Working agreements that age well: evidence over enthusiasm

I care about “Working agreements that age well” for one practical reason: it either leaves clearer ownership next week, or it was theater.

The collaboration tax is paid in ambiguity. Clarity is a kindness and a throughput strategy.

I will stay on this thesis — no adjacent manifesto, no tour of every neighboring discipline.

Shared roadmaps without shared capacity envelopes are fiction. Publish what each team can carry before you celebrate alignment.

Measure what you claim to care about. If Working agreements that age well only appears in kickoff slides and never in review, incident, or planning artifacts, it is branding. Put a verification signal where people already look.

A scene from delivery

Two teams “share a roadmap.” Neither owns the shared schema. Breaking changes collide. The repair is a schema owner, a deprecation clock, and an explicit consumer list — not a longer joint standup.

That scene is the thesis under calendar pressure. The response is not another status meeting — it is a control.

Mechanisms

Hero bridges — one person who translates between teams — feel invaluable until they burn out and the org discovers it had no interface. Document what the hero knows while they are still standing.

Treat conflict as design input. Resolve it in the interface, not in personality. Productive disagreement about SLOs and capacity envelopes beats political disagreement about who is more strategic.

Replace status karaoke with a short decision doc: context, options, choice, date, owner. Status belongs in a dashboard or a running doc; meetings are for conflict and irreversible choices.

The meeting you should replace with a doc is the one that only redistributes status. Keep the meeting that resolves contested decisions — and require options in writing before it starts.

Conflict as a design input means capturing the disagreement as constraints and options. Personality narratives are a dead end; interface narratives ship.

Interfaces between teams beat shared Slack channels. Channels are transport; interfaces are contracts. Confusing the two creates ambient anxiety and no ownership.

On Working agreements that age well, publish the interface: inputs, outputs, done criteria, and who gets paged. More Slack is not a substitute for a contract, and hero translators are a continuity risk.

Another texture

A cross-team launch checklist exists as a wiki page nobody updates. Convert it into a living interface doc with an owner and a last-reviewed date, or delete it.

Two orgs share on-call for a boundary service. Without a primary for each failure mode, pages bounce. Shared ownership without primaries is a page-routing bug.

Tradeoffs

Fewer meetings feel risky to managers who equate presence with control. Measure outcomes and escaped misunderstandings instead of seat time.

Written interfaces cost an afternoon. Unwritten interfaces cost a quarter. Choose deliberately.

Strict ownership can feel cold. Ambiguous ownership feels warm until the page. Prefer clear primary owners with named collaborators.

Keep the unit of progress small enough to finish under a full calendar: one owner clarification, one verification signal, one reversible control. Grand programs without weekly evidence become status machines.

Resist the urge to expand scope into neighboring slogans. If a control does not make Working agreements that age well more operable for the next person, leave it for another note. Dilution is how coherent essays become stitched scrapbooks.

Failure modes

Each of these is a missing control, not a personality problem:

• Meetings used as a substitute for written interfaces — everyone left “aligned,” nobody can explain ownership later.
• Hero bridges that mask missing contracts until attrition.
• Collaboration theater: more shared channels, same missing contracts.
• Roadmaps that look shared but hide capacity fiction — platform work scheduled as free.

What to do this month

• Quarterly: delete channels and meetings that produce no decisions or artifacts.
• After hand-off: verify the receiving team can operate the failure mode without a Slack ping to the sender.
• For every shared system: name a primary owner per failure mode.

Prefer async artifacts for continuity. Use meetings when the decision is contested or irreversible. Calendar density is not collaboration quality.

Joint incidents need a single incident commander even when two orgs are involved. Dual command is how pages bounce and customers wait.

Publish “done” definitions across teams before joint work starts — especially for platform and product hand-offs. Ambiguous done is how resentments form.

Hand-offs that preserve context include decision history, open risks, and the last failed assumption — not only a ticket ID. Context loss is a defect.

Working agreements that age well name how to change the agreement. Static culture docs become wallpaper within a quarter.

Write the boundary as an API: inputs, outputs, latency expectations, error semantics, and who gets paged. If you cannot write that page, you do not have alignment — you have hope.

Escalation paths that rely on knowing the right person are outages waiting for vacation. Publish who decides, by when, and what “stuck” means.

Product wants a feature; platform wants stability. Without a capacity envelope and SLO, every conversation becomes moral. With them, the conversation becomes scheduling and tradeoffs.

A working group creates a Slack channel for “alignment.” Six weeks later the channel is noise and decisions still live in side DMs. Delete the channel or convert it into a decision log with owners.

Close

Protocols beat vibes. Write the boundary, name the owner, make failure legible.

If someone new cannot explain Working agreements that age well from your artifacts in fifteen minutes, the practice is still private.